The objective of this study is to compare two classification methods pixel-based approach and object-based approach in mapping mangrove status in Can Gio biosphere reserve.
This study using SPOT-5 multispectral image with 10m spatial resolution, including Green, Red, Near infrared and SWIR bands which was accquired on February 24, 2011 in Can Gio District, Ho Chi Minh city for classifying and mapping the status of mangrove forest in Can Gio biosphere reserves by two methods pixel-based and object-based approach approach..The results showed that, with overall accuracy of 88.5% and Kappa index is 0,91, object-based approach has more accurate in comparison to pixel-based approach, which hasoverall accuracy is 83.1% and Kappa index is 0,89. Object-based approach has overcame several fundamental limitations of pixel-based approach as it is not divided objects into small fragments, especially when classify high resolution images, easy to perform vectorization of classification result. Howeverthis method requires a high level of image processing, such as the hardwares and softwares have to be powerful, more input information needed especially the knowledge of about the study area. Research shows that, object-based approach is an effectively method to classify multispectral images with high accuracy to mapthe current status of Can Gio mangroves forests.